A note on the perturbative properties of BPS operators

AEI-2002-070
CERN-TH/2002-232
LAPTH-935/02

b.ep-th/0209103

A note on the perturbative properties of BPS operators

Gleb Arutyunov*'**[], Emery Sokatchev* W

* Max- Planck- Institut fur Gravitationsphysik, Albert- Einstein- Institut,
Am Muhlenberg 1, D- 14476 Golm, Germany

* CERN Theoretical Division, CH-1211 Geneva 23, Switzerland
" Laboratoire d'Annecy-le-Vieux de Physique Theorique^ LAPTH
BP 110, F-74941 Annecy-le-Vieux Cedex, France

Abstract

We discuss the perturbative behavior of the 1/2 BPS operators in J\f = 2 SCFT on the
example of two very similar quadrilinear composite operators made out of hypermultiplets.
An explicit one-loop computation shows that one of them is protected while the other
acquires an anomalous dimension. Although both operators are superconformal primaries
in the free case, the quantum corrections make the latter become a 1/2 BPS descendant of
the Konishi multiplet, while the former remains primary. The comparative study of these
two operators at higher orders may be helpful in understanding the quantum properties of
the Konishi multiplet.

1 email:agleb@aei-potsdam.mpg .de
2 email:Emery . SokatchevOcern. ch

"On leave of absence from Steklov Mathematical Institute, Gubkin str.8, 117966, Moscow, Russia
3 UMR 5108 associee a l'Universite de Savoie

1 Introduction

A lot of progress in understanding the classical and quantum properties of the superconfor-
mal field theories has been achieved since the discovery of the AdS/CFT correspondence.
This includes both the development of the abstract superconformal representation theory
as well as the analysis of the correlation functions of concrete composite operators obtained
in different regimes of the correspondence.

A special class of composite operators in N = 4 SYM playing a privileged role in the
AdS/CFT dictionary are the so-called BPS short operators, i.e. operators annihilated by a
certain fraction of the Poincare supercharges Q. They form isolated series of representations
of the superconformal group with quantized (protected) conformal dimension (for a full
classification see, e.g., [[|).

It is important to realize that the symmetry protects such BPS operators only if they
are superconformal primary, i.e., if they are annihilated by all the generators S of special
conformal supersymmetry. For a given composite operator in N = 4 SYM the latter
property is not obvious. It may happen that an operator looks BPS short (it is annihilated
by some charges Q), but in fact it is a superconformal descendant of another, long multiplet.
At the same time, such an operator may still be a conformal primary (i.e., it is annihilated
by the conformal boosts) pure state and may have well-defined but unprotected conformal
dimension. The issue is rather subtle because the superconformal "primarity" is lost only
at the quantum level. This possibility was first pointed out in [pj in the M = 1 framework
and has since been reiterated many times (see, e.g., ||).

The fact that a long supermultiplet lying on the threshold of the unitary bound in the
classical theory is a reducible representation of the superconformal group has been known
for a long time (see, e.g., jl[]). Recently, a detailed analysis of the decomposition of long
multiplets into semishort or BPS short multiplets has been carried out in 0j. In particular,
when the superconformal primary long multiplet of the M = 4 (M = 2) algebra is a Lorentz
scalar, one finds 1/4 (1/2) BPS multiplets in its decomposition.

In this note we present a concrete and simple example of a "fake" and a "true" 1/2
BPS operators in N = 2 SCFT and discuss their nature in perturbation theory. After
recalling the necessary tools to define the Konishi superfield in the framework of the M = 2
superspace in Section 2, in Section 3 we extract from it a superconformal descendant (but a
conformal primary) having the quantum numbers and the appearance of an N = 2 1/2 BPS
operator. It is obtained by acting on the Konishi superfield with four (i.e., half of the total
number) spinor derivatives and by using the interacting field equations for the elementary
constituents. This operator is a single-trace quadrilinear combination of the elementary
hypermultiplet (HM) superfields and involves color group commutators]]] At the same time,
we construct another, very similar composite with fully symmetrized color indices which is
supposed to be a true protected 1/2 BPS primary operator. Our present understanding of
the short composite operators does not allow us to make definitive statements about the
latter, we can only rely on perturbative evidence j?]]. In Section 4 we examine the one-loop
corrections to the two-point functions of both operators and show explicitly the origin of

It was precisely the presence of a commutator term in the definition of a composite chiral operator that
was interpreted in |^ as a signal that it may be a descendant of some lower dimensional non-chiral operator.
For relatively low conformal dimensions this criterium for identifying descendant operators in perturbation
theory was used, e.g. in [^, [J.

1

the anomalous dimension of the former. Since this is precisely the anomalous dimension
of the "parent" Konishi multiplet, understanding the mechanism which distinguishes the
"fake" 1/2 BPS operator from the very similar "true" one at higher orders in perturbation
theory may prove useful in an attempt to compute the exact anomalous dimension of the
Konishi multiplet. In addition, the example that we consider in the paper can serve as a
concrete and simple illustration of the general representation theory discussion in Ref. {ij .
Finally, in Section 5 we formulate several open questions, the main one being: How to test
the "primarity" of a given operator?

2 M = 4 SYM in terms of M = 2 superfields

The two M = 2 ingredients of the M = 4 SYM theory are the HM and the J\f = 2 SYM
multiplet. In harmonic superspace M the former is described by a Grassmann (G-) analytical
(also frequently called "1/2 BPS") superfield of U(l) charge +1,

D+q+ = Dfq+ = q+ = q+(x,6 + ,0+ ,u) . (1)

Here D^^ = ufD* a A and uf, i = 1, 2 form an SU(2) matrix of harmonic variables on the
two-sphere S 2 ~ SU{2)/U{\). In the free case the HM satisfies the equation of motion

D ++ q + = , (2)

which takes the form of a harmonic (H-)analyticity condition. Here

D ++ = 8++ - 2i0 + a^§ + d lt (3)

is the supercovariant harmonic derivative in the G-analytic basis, which also plays the role
of the raising operator of SU (2). The solution of eq. (g) is the ultrashort on-shell superfield

q+ = q\x)ut + 9 +a ^ a {x) + 6+R«(x) + 2ie + a^+d fl f i (x) U r , (4)

where the physical components of the M = 2 HM, the doublet of scalar fields q l (x) and the
fermions ip, R, satisfy their free field equations

Uq^x) = <fhp = pR = .

The M = 2 SYM multiplet is described by another G-analytic superfield, this time
of 17(1) charge +2, V ++ (x, 9 + , 8 + , u). It plays the role of the gauge connection in the
covariantized harmonic derivative

V ++ = D ++ + igV ++ , (5)
so that the flat commutation relations with ^ are preserved:

[V ++ ,/J+J=0. (6)

2 Grassmann analyticity is the natural generalization of the notion of chirality in the case of extended
supersymmetry.

2

In the Wess-Zumino gauge the component content of V ++ is reduced to the off-shell N = 2
gauge multiplet:

V++ = -2i9 + a^6 + A^(x) - iV2(6 + ) 2 0(x) + iV2(9 1

) 2 (,

+ 3(6 + ) 2 (6 + ) 2 DV(x) U r u j . (7)

In the M = 4 theory the HM q + belongs to the adjoint representation of the gauge group
and interacts with the gauge multiplet through the usual minimal coupling:

\7++q+ = D ++ q + + ig [V ++ ,q + ] = 0. (8)

The harmonic gauge connection V ++ serves as the prepotential of the N = 2 SYM
theory. The derivatives do not need a connection in the frame where G-analyticity is
manifest (see @). The connections for the conjugate harmonic derivative V = D +
igV and for the spinor derivatives V~ A = + igA~^ are determined through the
conventional constraints

[V ++ ,V"~] =2D°, (9)

F~ s+J = v- 4 , (10)

where D° is the U(l) charge operator. Next, the field strength is defined by the relation

{V+ Vp} = 2ige a pW. (11)

It is covariantly chiral,
and harmonic independent,

VtW = V~W = 0, (12)

V++W = 0, (13)

as a consequence of the above constraints.

In the free case the curvature W satisfies the field equation

D+D + "W = . (14)

The coupling to N = 2 HM matter modifies this equation into

D+D+ & W = 8ig[q + ,q + ], (15)

where q + denotes a special conjugation on S 2 which preserves G-analyticity.

Eqs. (||) and ( |T5| ) are equivalent to the N = 4 SYM field equations written down in
terms of N = 2 superfields. They can be obtained from an off-shell action with manifest
N = 2 supersymmetry, which allows a straightforward quantization.

3

3 The Konishi multiplet and its 1/2 BPS descendant

The N = 2 Konishi super field is defined as follows:

K = Tr (hvW + g + V~~g + - q + V~~q + ^ = Tr + q i q l ) + 9 terms . (16)

Using the constraints (||) , (|l0|) , (|i~l]) , (|l2|) and the field equations (|8|) , ((l^) it is easy to check
that

D +a D+K = Ylig Tr (W[q + , q + }) (17)

and

D ++ K = 0. (18)

Equation ( |l7|) encodes the classical non-conservation of the Konishi current K = . . . +
9 l a^9ik^{x) + . . .. The H-analyticity condition (|l^) means that the chargeless superfield K
is harmonic independent

The quantum corrections lead to an anomaly Q in the right-hand side of eq. (|T^):

D +a D+K = Ylig Tr (W[q + , q + ]) + eg 2 D+D + °Tr (WW) , (19)

where c is a number. We can simplify eq. ( |l9| ) by hitting it with D^D +a . The quantum
anomaly term drops out; in the classical term the derivatives see only W and we can use
the field equation fli~5|). The result is

(D + ) 4 K = D^D +6i D +a D^K = -96g 2 Tr ([q+, q + ][t , q + })

= -192g 2 {Tr {q + q + q + q + ) - Tr (q + q + q + q + ) } . (20)

The left-hand side of this equations is annihilated by all of the G-analytic derivatives
D + ,D + , and so is the right-hand side, due to the G-analyticity of the HMs. Let us define
the single-trace composite operator of £7(1) charge +4 (equal to its naive dimension)

IC +A = Tr{[q + ,q + ][q + ,q + ]) ■ (21)

It is manifestly G-analytic and in addition is also H-analytic (see eq. (|5|)):

D+/C+ 4 = DlK +A = D ++ K +A = . (22)

Note the appearance of commutators under the trace in (|2l|), which indicates that the
operator has been derived through the use of the filed equation (|l5l).

A very similar operator, also both G- and H-analytic, is obtained by taking the sum in
the second line of eq. (p0|) :P1

+4 = Tr (q + q + q + q + ) + Tr (q + q + q + Q + ) ■ (23)

3 Note that the similar operator J = Tr (— W W + q + V~~q + — q + V~ - g + ) corresponds to one of the
Af = 2 supercurrent multiplets and satisfies the conservation conditions D +a D^J = D ++ J = even in the
presence of interaction.

4 There exist two other quadrilinear G- and H-analytic HM composites, Tr (g + g + g + g + ) and
Tr (g + g + g + q + ) , but they are not related to the Konishi multiplet, so we do not consider them here.

4

Despite the appearance, the two operators /C +4 and +4 are substantially different, but
this can only be seen in the interacting quantum theory. In the free case both operators
can be viewed as 1/2 BPS operators. In fact, the free Konishi superfield K$ of canonical
dimension two is a reducible representation of conformal supersymmetry (much like the
Konishi current k^{x) of canonical dimension three which is a reducible representation of
the conformal group). As such, it can be decomposed into several irreducible multiplets
(see 0] for details), one of which is the 1/2 BPS projection (D+) 4 K . Then the free
version of eq. (^) simply means that this 1/2 BPS projection vanishes. This statement is
manifestly super-Poincare invariant because [Q, (Z) + ) 4 ] = 0, but it is also superconformal,
since [S, (D + ) 4 ]Kq = provided that the superfield Kq has canonical dimension two. Then,
what remains in the free Konishi superfield Kq forms a semishort multiplet. As to the
right-hand side of eq. (p0|), in the free case it is not related to the Konishi multiplet. Still,
as long as it is made out of free HMs, ZC +4 undoubtedly is a superconformal primary 1/2
BPS multiplet, just like the other composite +i .

Now, it is well known that in the quantum interacting theory the Konishi superfield K
acquires an anomalous dimension |l^, ||, [Tl] ]. As soon as this happens, the left-hand side of
eq. (p0|) ceased to be superconformally covariant, since [S, (D + ) A ]K ^ if dim K > 2. In
this sense the 1/2 BPS projection (D + ) A K becomes a superconformal descendant of the now
long Konishi multiplet K. Nevertheless, this 1/2 BPS projection is still conformal primary
in the following sense: It can be checked that although the differential operator (D + ) does
not commute with the conformal boost generators K^, [K^ 1 , (D + ) 4 ] oc D + cr^D + D ++ , when
applied to the Konishi superfield this commutator still vanishes because of the H-analyticity
(or SU{2) irreducibility) condition fli~g|). The above discussion applies to the quadrilinear
composite operator fC +4 as well, since it is related to the 1/2 BPS non-covariant projection
of K through the field equation ( |20| ) . We conclude that in the interacting quantum theory
/C +4 should be regarded as a superconformal descendant (but at the same time conformal
primary) of the long Konishi multiplet.

On the other hand, there is no obvious reason why the second 1/2 BPS operator +4
should change its nature when the interaction is switched on. It does not seem to be related
via the field equations to any other, long multiplet, so it is likely to remain superconformal
primary. If this is true, its dimension should be protected by superconformal symmetry.
However, our present understanding of the making of composite operators does not give
us a direct and simple test for "primarity". The (non) preservation of this property in the
quantum theory is somehow related to the potentially singular nature of the composite
operators. It is an interesting problem to find out what exactly "goes wrong" with /C +4 but
not with C +4 . The best we can do at present is to make perturbative tests, see the next
section.

Note that in a similar manner we can construct 1/2 BPS operators of lower dimension
(charge). These are bilinear or trilinear in the elementary HMs constituents:

Of 2 = Tr (q + q + ) , 0+ 2 = Tr (q + q + ) , Of 3 = Tr {q + q + q + ) , = Tr (q + q + q + )

(24)

and conjugates. At this low dimension there can be no commutators under the trace,
therefore we can expect that the above operators are primary and thus protected. In
fact, the bilinears C +2 are current multiplets, they contain various components of the R
symmetry current of the N = 4 theory. This gives an additional reason why they should

5

be protected.

Another remark concerns the clear difference in the structure of the descendant JC +i
and its "parent" superfield K (|l6|). The latter is bilinear in the matter superfields q + , but
non-polynomial in the gauge superfield.0 However, its descendant /C +4 is polynomial in the
HMs and does not involve any gauge superfields. In this respect it strongly resembles the
(supposedly) primary operator +i (|2^) . We believe that the much simpler structure of
ZC +4 makes it a better candidate for further investigations than its parent K.

4 BPS operators in perturbation theory

As discussed in the preceding section, the quadrilinear composites ZC +4 and +4 , although
having the same properties of 1/2 BPS operators in the free case, are expected to behave
quite differently in the (perturbative) quantum theory. To get a better feeling of how this
happens, it is useful to reexamine the two-point functions of these operators at one loop.
Below we outline this calculation for /C +4 in the J\f = 2 harmonic superspace formalism ||.

To start with, Wick contractions with the HM Euclidean propagator (q + (l)q + (2))g = Q =
(12)/47r 2 xf 2 produce the free two-point function^

/r+4 , , r+4 , n 12iV 3 (iV 2 - 1) (12) 4

where (12) = it£ is the SU(2) invariant contraction of the harmonic variables at the
two points. Here and below we consider only the lowest component of /C +4 putting thereby
9 1 = 9 2 = 0.

At order g 2 we have graphs of two types, J\ and J2 (see Fig. 1).[] The graphs of the
type J\ are obtained by inserting a gluon propagator between any two (solid) HM lines.
The graphs of the type J2 always contain two HM loops connected by a gluon exchange.

5 In the harmonic formulation the non-linear object is the connection V whose expression in terms of
the prepotential V ++ via the constraint (^|) is non-polynomial Q; in the more familiar M = 1 formulation
the non-polynomial dependence is due to the gauge factor in Tr $e 19 ^$.

The color combinatorics is performed by using the relations Tr(T a T b ) = S ab , facdfbcd = 2N5 ab ,
famnfbnpfcpm = Nf a bc, where T" with a — 1, . . . N 2 — 1 are the SU(N) Lie algebra generators.

7 The self-energy contribution to the HM propagator vanishes in the Af — 2 formalism.

6

Jl J2

Figure 1. Two different topoiogies contributing to the two-point function of
Only the second graph is responsible for the anomalous dimension.

Explicitly, J\ and J 2 are given by the following analytic superspace integrals:

*W*S^M W Wc^m^W) (26)

x 12 J x 15 x 52 x 16 x 62

and

J2 „ <*!I / Wlif P|)(p (F++(5)F++(6)) . (27)

Here

J 12 J x 15 x 51 x 26 x 62

(y++(5)y++(6)) = (g5+ ~ ( f/ 6+)4 «5^(^,^ 6 )

is the propagator of the N = 2 prepotential F ++ in the Feynman gauge. With all external
9's set to zero, the super invariant coordinate difference entering the HM propagator between
points, e.g. 1 and 5 is defined as follows:

4, = (x l5 - 2i^-9+ae+ 5 ) = x 2 51 , (28)

(15)

where (15~) = uf l u^. The integration over the internal Grassmann and harmonic variables
is very easy and we find for J\ and J2 the following space-time expressions (dropping the
propagator factors):

J i ~ a i f — — ■> — — , J2 ~ / a \ ■ (29)

J x 15 x 16 x 56 x 25 x 26 J x 15 x 25

The integral in J\ is finite, therefore conformal invariance tells us that it is proportional to
l/x\ 2 - Thus Ji contributes only a contact term to the two-point function which vanishes for
separated points. The integral J2 is logarithmically divergent and computing it, for instance
in dimensional regularization, one finds a simple pole, J2 ~ 1/e. The residue determines
the anomalous dimension. Performing a multiplicative renormalization and doing carefully

7

the color combinatorics, we find for the renormalized operator ZC +4 the one-loop anomalous
dimension 7 = 3g 2 N / 47r 2 , i.e. precisely the same as for the renormalized Konishi operator.

This result is in agreement with the discussion in Section 3, were /C +4 was shown to arise
as a superconformal descendant of the Konishi multiplet. We clearly see that the 1/2 BPS
operator /C +4 , which was a superconformal primary in the free theory, looses this property
when the quantum corrections are turned on. Had it remained primary, its dimension would
have been protected by the superconformal symmetry.

In fact, this is what happens to the operator +4 . Repeating our one-loop calculation
for the two-point function of + , this time we find that only the graph J\ contributes while
J2 is absent due to the mismatch between the color antisymmetrization at the interaction
vertices and the symmetrization at the external points. However, as we have just seen, the
contribution of J\ is a contact term, so the two-point function of +A does not receive any
quantum correction at one loop (for non-coincident points) .0

Finally, applying the same arguments to the mixed two-point function (/C +4 C +4 ), we
again find that the graph J2 cannot contribute, whereas J\ still gives a contact term. Thus,
the two operators do not mix, at least at one loop. This situation is different from the
example in Ref. || concerning two dimension four operators in the 84 of SU(4). One of
them is a single-trace descendant, but the other is double-trace and they mix at one loop.

The above perturbative test gives indirect evidence that the operator C +4 remains
superconformal primary in the quantum theory, while /C +4 changes its nature. It would be
preferable to have another, direct way of testing "primarity".

Clearly, the descendant K, +A has the same anomalous dimension as its parent opera-
tor. Therefore, from the practical point of view, it might be easier to study the quantum
properties of K by instead examining ZC +4 . Indeed, /C +4 is an analytic gauge-invariant com-
posite which, contrary to K, does not involve any covariantizing gauge connection in its
definition. Moreover, the existence of a protected superconformal operator with the same
quantum numbers and with a very similar structure suggests important simplifications. It is
conceivable that those graphs which appear in the perturbative expansion of the two-point
functions for both +A and /C +4 should have a vanishing (or rather, contact term) contribu-
tion, as we have just seen in the simplest one-loop example. If this is true, a large subset of
graphs could be eliminated from the calculation of the anomalous dimension of /C +4 . This
point, however, requires further study because at higher loops the protection mechanism for
+A might be more involved, perhaps not due to the convergence of individual graphs but
rather to non-trivial color combinatorics leading to cancellation of individual divergences.
In any case, the question of understanding how an operator like +A might be protected to
all orders in perturbation theory is still open and certainly very interesting.

5 Conclusions

In this note, in the context of J\f = 2 SCFT we have demonstrated the different perturbative
behavior of operators having the same quantum numbers of a 1/2 BPS state. We considered
the simplest example of two analytic superfields composed of four HMs. Both of them
are superconformal primary operators in the free theory but when the quantum effects

8 This result is stronger than just saying that the operator +A has a protected dimension. In fact, this
is what is believed to happen to all 1/2 BPS primary operators [ [l2[ .

8

are turned on, the first one ceased to be primary with respect to the special conformal
supersymmetry and becomes a member of the long unprotected Konishi multiplet. The
second one remains a superconformal primary at the quantum level and hence is protected.

As argued in Ref. Q, in some special cases one is able to tell that a short or a semishort
operator with given quantum numbers cannot possibly be contained as a descendant in an-
other, long multiplet, and should thus be automatically protected. Our examples are clearly
not of this type. At present, we do not have a better way to decide whether one or the other
of our 1/2 BPS operators could acquire an anomalous dimension but making perturbative
tests. In this context we remark that the various "non-renormalization theorems" for short
operators in the literature have always been based on the (implicit) assumption of "pri-
marity" . As we have tried to show in this note, demonstrating the "primarity" of a given
composite short operator at the quantum level seems equally difficult, in practice equivalent
to proving that it is protected. It would be highly desirable to find some intrinsic criterion
for determining the superconformal properties of an operator in perturbation theory.

The discussion above raises the intriguing question about what precisely is responsible
for the loss of "primarity" of certain operators. Progress in this direction might lead to
explaining the mechanism which protects short operators, on the one hand. On the other, it
may prove helpful in revealing the deep origin of the anomalous dimension of long multiplets
and eventually to its exact evaluation.

Our discussion also applies to the so-called BMN operators recently introduced in the
context of the SYM/plane-wave correspondence [jDj. Indeed, a BMN operator with impu-
rities is a mixture of operators in different 5'C/(4) representations. It contains, in particular,
irreducible SU (4) pieces which are single-trace operators involving commutators of elemen-
tary fields. Thus, some of them might be superconformal descendants of lower-dimensional
operators. Since the span of dimensions in a supermultiplet is finite, it is conceivable that
in the BMN limit of infinite quantum numbers the difference between a primary field and
its supersymmetry descendant actually disappears or becomes insignificant. It would be
interesting to better understand this point.

Acknowledgements

We are grateful to C. Callan, S. Ferrara, P. Howe, H. Osborn and E. Sezgin for many
useful discussion. G. A. was supported by the DFG and by the European Commission RTN
programme HPRN-CT-2000-00131, and in part by RFBI grant N99-01-00166.

9

References

[i]

V. K. Dobrev and V. B. Petkova, "All Positive Energy Unitary Irreducible Representations Of
Extended Conformal Supersymmetry," Phys. Lett. B 162 (1985) 127.

P. Heslop and P. S. Howe, "On harmonic su perspaces and sup erconformal fields in four dimen-
sions," Class. Quant. Grav. 17 (2000) 3743, |hcp-th/0005135| .

S. Ferrara and E. Sokatchev, "Super conformal interp retation of BPS states in AdS geometries,"
Int. J. Theor. Phys. 40 (2001) 935, |hep-th/0005151 .

[2]

M.

E. Witten, "Aii ti-de Sitter space and holography," Adv. Theor. Math. Phys. 2 (1998) 253,
ticp-th/980215C .

S. Ferrara a nd A. Zaffaroni, "Bulk gauge fields in AdS supergravity and supersingletons," |hcp-
th/980709C .

L. Andrianopoli, S. Ferrara, E. Sokatchev and B. Zupnik, "Shortening of primary operators in
N-ext cndcd SCFT(4) a nd harmonic-superspace analyticity," Adv. Theor. Math. Phys. 3 (1999)
1149, |hcp-th/99 12007.

[4]
[5]

F. A. Dolan and H. Osborn, "On Short and S emi-Short Representations for Four Dimensional
Superconformal Symmetry," hep-th /0209056[

M. Bianchi, S. Kovacs, G. Rossi a nd Y. S. Stancv, "On the logarithmic behavior in N = 4 SYM
theory," JHEP 9908 (1999) 020, |hcp-th/9906l"8S|; "Prope rties of the Konishi multiplet in N =
4 SYM theory," JHEP 0105 (2001) 042, [hcp-th/0104016 .

[6] A. V. Ryzh ov, "Quarter BPS operators in N = 4 SYM," JHEP 0111 (2001) 046, hep

th/0109064; E. D'Hoker and A. V. Ryzh ov, "Three-point functions of quarter BPS operators
in N = 4 SYM," JHEP 0202 (2002) 047, |hcp-th/0109065.

[7] E. D'Hoker, D. Z. Frccdman and W. Skiba, "Field theory tests for c orrelators in the AdS/CFT

correspondence," Phys. Rev. D 59 (1999) 045008, |hcp-th/9807098
S. Penati, A. Santambrogio and D. Zanon, "Two ^pomt functions of chiral operators in N =
4 SYM at order g**4," JHEP 9912 (1999) 006, |hcp-th/9910"l97| ; "More o n correlators and

contact terms in N = 4 SYM at order g**4," Nucl. Phys. B 593 (2001) 651, |hcp-th/0005"223

[8]

A. Galpcrin, E. Ivanov, S. Kalitsyn, V. Ogievetsky and E. Sokatchev, "Unconstrained N=2
Matter, Yang-Mills And Supergravity Theories In Harmonic Superspace," Class. Quant. Grav.
1 (1984) 469.

A. S. Galperin, E. A. Ivanov, V. I. Ogievetsky and E. S. Sokatchev, "Harmonic Superspace,"
Cambridge Univ. Press (2001) 306 p.

[9]

[10] D. Ansclmi, "Q uantum conforma l algebras and closed conformal field theory," Nucl. Phys. B

K. Konishi, "Anomalous supersymmetry transformation of some composite operators in
SQCD," Phys. Lett. B135, 5 (1984) 439.

554 (1999) 415, hcp-th/9811149

[111

[12]

[13]

G. Arutyunov, S. Frolov and A. Petkou, "Perturbative and instanton correctio ns to the OPE of
CPOs in N = 4 SYM(4)," Nucl. Phys. B 602 (2001) 238, E B 609 (2001) 540, |hcp-th/0010137.

D. Z. Freedman, S. D. Mathur, A. Matusis and L. Rastelli, 'Co rrelation function s in the
CFT(d)/AdS(d+ 1) correspondence," Nucl. Phys. B 546 (1999) 96, |hcp-th/9804058 :
P. S. Howe, E. Sokatchev and P. C. West, "3-point functions in N = 4 Yang- Mills," Phys. Lett.
B 444 (1998) 341, |hcp-th/9808"l62 .

D. Berenstein, J. M. Maldacena and H. Nastase, "Strings in flat space and pp waves from N
4 super Yang Mills," JHEP 0204 (2002) 013, ^ep-th/0202021 .

10